This study introduced advanced Artificial Intelligence (AI) tools to pipeline integrity management, enhancing the accuracy of defect detection and improving risk-based decision-making. Using in-line inspection (ILI) records and field survey data, machine learning models with data augmentation were employed to detect anomalies by identifying corrosive soil environment at an early stage. The results show that pipeline corrosion was more likely to occur in soil environments with high moisture content, low pH, and high redox potential. Bayesian Neural Networks (BNN) were utilized to predict the evolution of corrosion defects, which can be used to calculate the failure probability of pipelines in the future. Corrosion prediction uncertainty increases significantly in the long term due to the dynamic and stochastic nature of corrosion growth. The study results show the benefits of AI-based approaches to enhance pipeline integrity management.
Cui et al. (Thu,) studied this question.